Cybersecurity Policy Report, Oversight Checklist Released for EU Institutions’ ADM Tools, (Jul 6, 2026)
By Tony Foley
A self-assessment checklist outlining the issues to be considered by European Union institutions, bodies, offices, and agencies (EUIs) in considering measures taken to establish human oversight of automated decision-making (ADM) systems was released last week by the European Data Protection Supervisor (EDPS).
The checklist defines “human oversight” as a set of measures aimed at minimizing potential negative effects on individuals during and after the operation of ADM systems. It is often proposed as a key measure for mitigating risks associated with the deployment of such systems, but EDPS emphasized that human oversight was not a “silver bullet” capable of fully eliminating the risks inherent in such systems, including biases that influence human reactions with ADM systems and the circumstances under which a system may operate.
The checklist is broken into two sections, the first including controls that apply broadly to any system whose decisions or recommendations may impact individuals, covering governance, user training, accountability, and explainability, among others, and the second applicable when there are operators responsible for supervising ADM systems and have the authority to intervene in their actions, including determining the proper authority for intervention, timelines for review, interface designs and regular internal and external monitoring.
News: InternationalLegislation DataPrivacy AINews